Space Robot Vision Calibration with Reference Objects from Motion Trajectories

For lack of reference objects in space, an on-orbit vision calibration method is proposed based on the precise localization of manipulator end of space robot, which generates reference objects through programming motion trajectories of the manipulator. Firstly, a general framework of the method is introduced. Secondly, error analysis is applied to the calibration method. According to the prior knowledge that the calibration error of the chessboard target is linear to the image coordinate error of the control point, the error transfer from localization of manipulator end to the image coordinate of the control point is derived, and an equivalent error of the image coordinate of the control point is further derived by the combination of the two kinds of errors. Finally, the linear relationship between the calibration error and the equivalent error is validated by simulation experiments, and the effects of reference object size, relative distance, control point density and structure factors on the accuracy of the proposed vision calibration method are analyzed.